Chapter 1.

Introduction
Spiral antennas, a type of frequency independent antenna, have been studied for
over 40 years. Frequency independent antennas provide uniform electrical characteristics
over a wide frequency band. However, frequency independent antennas typically have
broad radiation patterns and low gain, which is not suitable for many applications. One
method to overcome this limitation is to use an array of frequency independent antenna
elements. This approach allows for pattern control and higher gains, but the wideband
characteristics of the frequency independent element are lost in the array environment.
Inter-element spacing usually limits array bandwidth to a value much less than a
frequency independent element can achieve outside an array.
A unique type of frequency independent antenna element, the star spiral, is
presented in this dissertation. The star spiral is a type of slow-wave spiral that works
very well in the wideband array with variable element sizes (WAVES) due to its unique
shape. The WAVES array concept has been presented in the literature to a limited
degree. This dissertation will further analyze the WAVES array concept and present
designs using the star spiral in a WAVES array.

1.1 Motivation

Conventional antenna arrays are typically bandwidth limited by the individual

antenna element at the lower frequency and by the formation of grating lobes due to
inter-element spacing at the higher frequency. Techniques such as unequally spaced
arrays, shared aperture arrays, and fractal arrays have been used to improve array
bandwidth. Stutzman proposed another technique, called the wideband array with
variable element sizes (WAVES) (Stutzman, 1985). It is a shared aperture technique that
uses different element sizes to cover each octave of desired bandwidth. Frequency
independent antenna elements, specifically the Archimedean spiral, were chosen because
of their wide bandwidth. For each consecutively higher octave of bandwidth, the
diameter of the Archimedean spiral is halved, allowing for tighter array spacing. Only
the largest elements are active over the first octave of bandwidth. For the next octave, the
two largest element sizes are active which maintains the array spacing less than one

1
wavelength. In theory the process can continue to higher octaves. Three octave designs
have been proposed.
Much of the previous work on WAVES was theoretical with some measurements.
However, for a simple linear WAVES array, simulations show that there is a performance
gap between the low and high octaves. An investigation of slow-wave spiral antenna
techniques was conducted to bridge the gap between the two octaves. The star spiral
antenna is the result of that study. The unique shape of the star spiral allows for tighter
element packing in a WAVES array and combined with its slow-wave nature the star
spiral is able to eliminate the performance gap in a linear WAVES array. The shape of
the star spiral is also advantageous in planar array packing and reducing blockage in
overlapping arrays.

1.2 Literature Review

1.2.1 Wideband Arrays

The effect of inter-element spacing is an inherent problem in designing wideband
arrays. It is desirable for inter-element spacing to be small, but the minimum spacing is
limited by antenna element size and mutual coupling effects. As the frequency of
operation is increased the inter-element spacing increases in terms of wavelength, and
grating lobes appear for a spacing of one wavelength. Grating lobes are typically
undesirable so the array bandwidth is limited by element size at the low frequency and
the formation of grating lobes at the upper frequency. For a typical array this frequency
range may correspond to an inter-element spacing between a half wavelength and one
wavelength or a 2:1 bandwidth. Antenna elements, such as spirals, can have bandwidths
of 9:1 or greater so techniques for increasing array bandwidth are very valuable.
Three of the main techniques for increasing array bandwidth are unequal array
spacing, shared apertures, and non-rectangular array geometries. Much of the unequal
array spacing work was done in the early 1960’s. Unz (1960) studied a linear array with
arbitrarily distributed elements. He found that the additional degree of freedom created
by the random distribution of elements allowed him to achieve the same performance of
an equally spaced array with fewer elements. The effects of various unequal spacing

2
schemes, such as logarithmic spacing, non-monotonically increasing spacing, and
elimination of multiple spacing, were investigated by King, et al., (1960). King found
that compared to an equally spaced array fewer elements were required for a desired
bandwidth and grating lobes were replaced by sidelobes.
A method for quantifying the bandwidth of an unequally spaced array was
presented by Bruce (1962). The sensitivity of the array pattern is found by taking the
partial derivative of the pattern function with respect to the wavenumber. The most
broadband arrays occur when the change in the partial derivative is stationary or the
second partial derivative is zero. The second partial derivative reduces to the determinate
of Bessel functions. The closer the determinant is to zero the more broadband the array
pattern becomes.
Theoretical models for designing and analyzing unequally or randomly spaced
arrays were introduced by Ishimaru (1962) and Lo (1964). Ishimaru showed that by
using Poisson’s sum formula and a new function he called the source position function,
unequally spaced arrays with a desired radiation pattern could be designed. Ishimaru’s
technique was also valid for designing arrays on curved surfaces. Lo used a probabilistic
approach to analyze large randomly spaced arrays. He found that sidelobe level was
closely linked to the number of elements in the array, but very weakly connected to
aperture size. Lo also found that directivity was proportional to number of elements and
beamwidth was related to aperture size. The implication of Lo’s study was that
beamwidth could be changed dramatically by lengthening the array with little effect on
sidelobe level.
Skolnik (1964) used a computer technique known as dynamic programming to
optimize a thinned, unequally spaced array. The array he designed consisted of 25
elements in a 50-wavelength aperture. Another method for designing a thinned and
broadband array using unequal array spacing was shown by Ishimaru and Chen (1965)
that used Anger functions to express the array pattern. They were able to design an array
with the desired beamwidth and sidelobe level for an average element spacing of many
wavelengths. Further work was done by Bratkovic (1973) on optimizing unequally
spaced arrays with dynamic programming. He found that the optimum array had
monotonic spacing.

3
 The use of randomly spaced subarrays as elements in a uniform linear array has
also been investigated (Goffer, et al., 1994). Probabilistic expressions for grating lobe
location and level and gain loss are given. The use of random subarrays allows for better
grating lobe performance compared to a standard uniform array. A paper by Yu (1997)
presented a method for reducing the sidelobes of an unequally spaced array by spacing
perturbation. Least square error approximation and point matching were used with the
radiation pattern to determine the appropriate element locations. A small amount of
beamwidth broadening was also found. Another paper shows that by using a density
tapering method with a radial warping function and minimum spacing greater than a
wavelength an array with no grating lobes can be achieved (Anderson, et al., 1998).
A recent paper by Tripp and Papanicolopoulos (2000) describes a wideband array
geometry designed using frequency independent antenna techniques. The array is called
the log-periodic phased array. The array grid is defined by angle with a log-periodic
relationship along each radial line. The structure looks concentric rings of elements
where element size and array spacing increases with the radius of the array. Another type
of frequency independent antenna array is the fractal array. A fractal array is recursively
generated using fractional dimensions of the previous iteration. Currently fractal
antennas and arrays are very popular in the literature. A review of fractal arrays is given
by Werner, et al. (1999).
The previous papers dealt mainly with increasing array bandwidth using
unequally spaced antenna elements of uniform size. Other techniques, such as shared
aperture and non-rectangular grids, also can improve array bandwidth. A shared aperture
array containing elements in L, S, and C frequency bands has been built and measured
(Boyns and Provencher, 1972). Mutual coupling arrays of this type can be a problem, but
this problem can be lessened by using elements with different polarizations. A 96-
element phased array has been designed to operate over a one-octave bandwidth
(Laughlin, et al., 1972). Each row of elements is staggered from the previous row
creating a triangular grid of elements. A similar array was designed by Robinson, et al.
(1989). The array consisted of two offset linear arrays of quadridged waveguide horns.
A bandwidth of approximately 2.5:1 was achieved.

4
1.2.2 Spiral Antennas
Spiral antennas belong to a class of antennas known as frequency independent
antennas. An antenna that can be completely specified by angles is frequency
independent (Rumsey, 1957). The impedance and pattern performance of a frequency
independent antenna are constant with frequency. In practice, an antenna must have a
finite size so it only exhibits frequency independent behavior over a certain frequency
range. A practical spiral antenna exhibits frequency independent behavior over a
frequency range determined by its inner and outer radius.
Much of the early work on spiral antennas was published in the late 1950’s and
early 1960’s. The planar equiangular spiral antenna and the unidirectional equiangular
spiral or conical log spiral antenna were presented by Dyson (1959a, 1959b).
Bandwidths of greater than 20:1 were observed with nearly constant impedance and
pattern performance. Bawer and Wolfe (1960) collected much of the previous work on
spiral antennas and summarized the performance of the spiral antenna for variations in
different parameters. They looked at the equiangular, Archimedean, and square spiral
antennas.
Much of the early work on spiral antennas was based on experiment and the band
theory. The band theory essentially means that the spiral operates in the region where the
circumference of the spiral is equal to a wavelength. In the early 1960’s more rigorous
mathematical explanations were pursued. Curtis (1960) derived the radiation patterns for
an Archimedean spiral by approximating the spiral as a series of semicircles. Wheeler
(1961) looked at the radiation from various regions of an equiangular spiral using a
similar technique to Curtis, but without the semicircle approximation. A more general
explanation of the spiral antenna was given by Cheo, et al., (1961). They solved
Maxwell’s equations for an equiangular spiral consisting of an infinite number of
coplanar arms. It was found that the frequency independent behavior of the spiral is
dependent on the curvature of each arm of the spiral.
Spiral antennas are typically backed by a lossy cavity. The lossy cavity improves
the low frequency impedance behavior and axial ratio of the spiral by reducing
reflections from the end of the each arm of the spiral. The lossy cavity also absorbs the
back radiation from the spiral providing for a larger pattern bandwidth by reducing the

5
reflection from the ground plane that causes pattern nulls. Spirals with bandwidths of 9:1
or greater are common. The bulk of the lossy cavity and the gain reduction due to the
loss are the two major drawbacks of using a cavity-backed spiral. As a result, conductor
backed spirals have gained some popularity for certain applications. Experiments have
shown a 1.2:1 circular polarization bandwidth for conductor backed spiral antennas
(Nakano, et al., 1986 and Wu, 1994). For very low profile designs a slot spiral can be
used to further reduce the height of the antenna. Slot spirals with bandwidths of 25:1 and
depths of one hundredth the wavelength of the lowest frequency have been demonstrated
(Nurnberger and Volakis, 1999).
Size reduction of spiral antennas has been studied for many years. Material
loading is one way to reduce the size but material loss and weight can be a problem in
some applications. Slow-wave spiral techniques were developed to overcome the
problems inherent in material loading. Adding some type of high frequency profile, such
as a zigzag or sine wave, to the spiral and increasing the circumference of the spiral, such
as the square spiral, are ways of producing a slow-wave spiral. The radiation zone for a
specific wavelength is moved closer to the center of the spiral when slow-wave
techniques are employed. This effectively reduces the velocity of propagation along the
length of the spiral, which reduces the low frequency cutoff of the spiral providing for
size reduction (Roland and Patterson, 1967). The low frequency cutoff may also be
reduced by resistively terminating the end of each arm of the spiral to reduce reflections
from the end of the spiral, but this reduces efficiency, and, thus, gain.

1.2.3 Wideband Array With Variable Element Sizes

Stutzman (1985) first introduced the wideband array with variable element sizes
(WAVES) concept. He presented the theory and performed a feasibility study by
measuring the patterns of an Archimedean spiral with copper disks layered above the
spiral representing higher octave antennas. A two-octave, eight element, planar array of
Archimedean spirals was built and tested by Shively and Stutzman (1988, 1990). The
planar array was measured along the diagonal where a triangular lattice and amplitude
taper improved the radiation pattern performance of the array. The smaller spiral
elements were switched on when the grating lobe due to the larger elements became too

6
large. Other theoretical work has been done by Chatzipetros (1993). He used array
theory to predict the performance of some three-octave WAVES array designs. All of the
previous work on WAVES has consisted of array theory predictions and measurements of
the Shively eight element, planar array.

1.3 Dissertation Overview

This dissertation is organized into three main parts. The first part is contained in the
first three chapters and is primarily a review of previous work and other background
materials. Chapter 1 is a review of the relevant literature. The basic theory and some
simple simulations of an Archimedean spiral are presented in Chapter 2. A verification
of the NEC4 code for simulation of Archimedean spirals is also performed (Burke, 1992).
NEC4 is used as the primary simulation tool due to its relatively fast run times and
versatility in regards to geometry input. A circular Archimedean spiral will be built and
measured to validate the simulation tools and to be used as a comparison with the star
spiral in later chapters. Past work on the wideband array with variable element sizes is
summarized in Chapter 3. Linear and planar WAVES arrays of Archimedean spirals are
simulated and the need for a slow-wave spiral is detailed.
The next section consists of a detailed analysis of the star spiral antenna element
in Chapter 4. The star spiral was developed to overcome the performance gap in the
linear WAVES array of Archimedean spirals presented in Chapter 3. The star spiral also
yields itself to some interesting array geometries and can improve array packing.
Simulations of the star spiral in an array environment are shown Chapter 5. The input
impedance and scan performance of the star spiral will be investigated using Analysis
Software for Infinite Arrays (ASIA) (LaPean, 1996). The last part of the thesis details
the use of the star spiral in WAVES arrays. Improvements in performance, a 3-octave
array, scan performance, and various array architectures will all be presented in Chapter
5. Finally, conclusions and future work will be given in Chapter 6.